Artificial intelligence research may have hit a dead end

Artificial intelligence research may have hit a dead end (from 2021, so possibly out of date?)

Thomas Nail

Furthermore, the cause and pattern of neuron firing are subject to what neuroscientists call "spontaneous fluctuations." Spontaneous fluctuations are neuronal activities that occur in the brain even when no external stimulus or mental behavior correlates to them. These fluctuations make up an astounding 95% of brain activity while conscious thought occupies the remaining 5%. In this way, cognitive fluctuations are like the dark matter or "junk" DNA of the brain. They make up the biggest part of what's happening but remain mysterious.   

What if noise is the new signal? What if these anomalous fluctuations are at the heart of human intelligence, creativity, and consciousness? This is precisely what neuroscientists such as Georg Northoff, Robin Carhart-Harris, and Stanislas Dehaene are showing. They argue that consciousness is an emergent property born from the nested frequencies of synchronized spontaneous fluctuations. Applying this theory, neuroscientists can even tell whether someone is conscious or not just by looking at their brain waves. 

AI has been modeling itself on neuroscience for decades, but can it follow this new direction? Stanislas Dehaene, for instance, considers the computer model of intelligence "deeply wrong," in part because "spontaneous activity is one of the most frequently overlooked features" of it. Unlike computers, "neurons not only tolerate noise but even amplify it" to help generate novel solutions to complex problems.  

Artificial intelligence research may have hit a dead end (from 2021, so possibly out of date?)

Thomas Nail

It doesn't look to me that this research (if valid at all) would lead to a dead end in the current trends in AI. The big thing in current AI especially generative AI development is large language processor systems like ChatGPT4, that logically process and use as data sources untold millions of lines of human-produced Internet information and use statistical methods to come up with responses (or it can be other types of human-produced information, or medical data, etc. depending on the intended purpose of the AI system. The human-like responses of some of these AI systems are based on analyzing vast amounts of human-produced verbiage or other intelligent input, and it doesn't appear as of yet that anything like truly conscious general intelligence AI will emerge from such systems. Exactly how human consciousness comes about in the first place doesn't seem to be very relevant to this.

Of course, if this new research were actually on the right track, then it would offer another perhaps more genuine and effective way of actually generating artificial consciousness in an artificial general intelligence system, which would also be vastly more dangerous to humanity. 

However, there is a great amount of evidence (as discussed in this forum, especially from certain paranormal phenomena) that strongly militate against such a materialist neuroscientific attempt to explain consciousness. To say nothing of all the strong philosophical arguments against such a physicalist position.

It doesn't look to me that this research (if valid at all) would lead to a dead end in the current trends in AI. The big thing in current AI especially generative AI development is large language processor systems like ChatGPT4, that logically process and use as data sources untold millions of lines of human-produced Internet information and use statistical methods to come up with responses (or it can be other types of human-produced information, or medical data, etc. depending on the intended purpose of the AI system. The human-like responses of some of these AI systems are based on analyzing vast amounts of human-produced verbiage or other intelligent input, and it doesn't appear as of yet that anything like truly conscious general intelligence AI will emerge from such systems. Exactly how human consciousness comes about in the first place doesn't seem to be very relevant to this.

Of course, if this new research were actually on the right track, then it would offer another perhaps more genuine and effective way of actually generating artificial consciousness in an artificial general intelligence system, which would also be vastly more dangerous to humanity. 

However, there is a great amount of evidence (as discussed in this forum, especially from certain paranormal phenomena) that strongly militate against such a materialist neuroscientific attempt to explain consciousness. To say nothing of all the strong philosophical arguments against such a physicalist position.

Finding the proper structure of "physical stuff" to instantiate consciousness is a metaphysically neutral position IMO, not a physicalist one. We know some arrangement of "physical stuff" correlates with consciousness, though whether this is enough for an instantiation of a first person POV would have to wait for an experiment...not that said experiment would immediately tell us which metaphysics is correct...

But for our immediate concerns I think the question of how human creativity works in regards to even just correlates neuroscience would find puts a damper on the idea that we can magically make AI better by just pouring loads of data into it. 

If there had been more consideration for the potential value of indeterministic "noise" in the brain contributing to human tasks we may not have had these issues with driverless cars being public experimentation for corporate benefit, and also could stave off claims that AI art generation is doing what humans do rather than just stealing from real artists.

Finding the proper structure of "physical stuff" to instantiate consciousness is a metaphysically neutral position IMO, not a physicalist one. We know some arrangement of "physical stuff" correlates with consciousness, though whether this is enough for an instantiation of a first person POV would have to wait for an experiment...not that said experiment would immediately tell us which metaphysics is correct...

But for our immediate concerns I think the question of how human creativity works in regards to even just correlates neuroscience would find puts a damper on the idea that we can magically make AI better by just pouring loads of data into it. 

If there had been more consideration for the potential value of indeterministic "noise" in the brain contributing to human tasks we may not have had these issues with driverless cars being public experimentation for corporate benefit, and also could stave off claims that AI art generation is doing what humans do rather than just stealing from real artists.

Metaphysically neutral? This is a materialist notion of the nature of consciousness, to the core. It says that neuronal activity in the brain generates synchronized spontaneous fluctuations which in turn generate consciousness as an emergent property; consciousness is not an immaterial but still correlating property.

"They argue that consciousness is an emergent property born from the nested frequencies of synchronized spontaneous fluctuations. Applying this theory, neuroscientists can even tell whether someone is conscious or not just by looking at their brain waves."

I think this makes it clear that this is a physicalist neuroscience attempt to explain consciousness materialistically. It rules out the existence of any sort of immaterial spirit or soul independent of the physical brain - if the brain is dead there are no "sychronized spontaneous fluctuations" arising from neuronal activity, and consciousness is wiped out since it is a product of neural activity.

Metaphysically neutral? This is a materialist notion of the nature of consciousness, to the core. It says that neuronal activity in the brain generates synchronized spontaneous fluctuations which in turn generate consciousness as an emergent property; consciousness is not an immaterial but still correlating property.

"They argue that consciousness is an emergent property born from the nested frequencies of synchronized spontaneous fluctuations. Applying this theory, neuroscientists can even tell whether someone is conscious or not just by looking at their brain waves."

I think this makes it clear that this is a physicalist neuroscience attempt to explain consciousness materialistically. It rules out the existence of any sort of immaterial spirit or soul independent of the physical brain - if the brain is dead there are no "sychronized spontaneous fluctuations" arising from neuronal activity, and consciousness is wiped out since it is a product of neural activity.

The generation of consciousness as an emergent property is an interpretation of data, the spontaneous fluctuations are the data.

The generation of consciousness as an emergent property is an interpretation of data, the spontaneous fluctuations are the data.

An emergent property is tied to the operation of the system from which it emerges. In this case the theory is that the system from which consciousness emerges is the actions of billions of neurons in the brain. The actions of these neurons cease at death, therefore the emergent property (consciousness) ceases and no longer exists. So in this theory consciousness is totally dependent on the workings of a functional brain. So this theory is a thoroughgoing materialist neuroscientific "explanation" of consciousness. That is hardly "metaphysically neutral". 

Simple chemistry/physics analogy: "wetness", liquidity, and a skin tension effect are emergent properties of H2O molecules at a certain temperature and pressure and with a suitable containment.  These properties emerge from the ultimately quantum mechanical behavior of these molecules when in close contact. Take away the assemblage of H2O molecules or the necessary pressure or their quantum mechanical properties and there is no wetness or liquidity, etc.

An emergent property is tied to the operation of the system from which it emerges. In this case the theory is that the system from which consciousness emerges is the actions of billions of neurons in the brain. The actions of these neurons cease at death, therefore the emergent property (consciousness) ceases and no longer exists. So in this theory consciousness is totally dependent on the workings of a functional brain. So this theory is a thoroughgoing materialist neuroscientific "explanation" of consciousness. That is hardly "metaphysically neutral". 

Simple chemistry/physics analogy: "wetness", liquidity, and a skin tension effect are emergent properties of H2O molecules at a certain temperature and pressure and with a suitable containment.  These properties emerge from the ultimately quantum mechanical behavior of these molecules when in close contact. Take away the assemblage of H2O molecules or the necessary pressure or their quantum mechanical properties and there is no wetness or liquidity, etc.

Even if the authors explicitly state they are seeking a materialist theory, the recorded data remains metaphysically neutral -that's just the recording of fluctuations in neuronal firing. Regardless of one's metaphysical position, one has to explain why the brain exists and behaves like it does. I don't see how an Idealist, Dualist, Neutral Monist, Panpsychist, etc are able to complete their metaphysical picture without explaining what the brain is doing and its relation to consciousness.

Additionally, without picking a metaphysics, we can observe the structure of the brain - and by structure I am including processes in time along with its form in space. It's "spontaneous fluctuations" that happen in the brain, which make it structurally different from Turing Machines we produce through computer engineering. Why, regardless of metaphysics, we have reason to doubt machine "learning" is going handle tasks the same way our brains + our consciousness does.

Even if the authors explicitly state they are seeking a materialist theory, the recorded data remains metaphysically neutral -that's just the recording of fluctuations in neuronal firing. Regardless of one's metaphysical position, one has to explain why the brain exists and behaves like it does. I don't see how an Idealist, Dualist, Neutral Monist, Panpsychist, etc are able to complete their metaphysical picture without explaining what the brain is doing and its relation to consciousness.

Additionally, without picking a metaphysics, we can observe the structure of the brain - and by structure I am including processes in time along with its form in space. It's "spontaneous fluctuations" that happen in the brain, which make it structurally different from Turing Machines we produce through computer engineering. Why, regardless of metaphysics, we have reason to doubt machine "learning" is going handle tasks the same way our brains + our consciousness does.

Then I guess the writer is guilty of bad reasoning and bad philosophy in apparently assuming that with his theory he has validly proposed a materialist emergentist neuroscience explanation of consciousness. As you point out, he has not completed his theory by accounting for at least one critical thing in the total picture. Dualists and Idealists at least provisionally have such explanations.

Then I guess the writer is guilty of bad reasoning and bad philosophy in apparently assuming that with his theory he has validly proposed a materialist emergentist neuroscience explanation of consciousness. As you point out, he has not completed his theory by accounting for at least one critical thing in the total picture. Dualists and Idealists at least provisionally have such explanations.

Well Nail leans materialist but he's said it doesn't matter that much to him if people say his theory is panpsychist. But yes anyone who thinks something that has no mental character, like matter, is going to generate consciousness is grasping at straws & ignoring the Something from Nothing problem.

That said, I do think Nail is on to something where he says indeterminacy that is neither random nor deterministic has an important role to play in consciousness, the formation of the physical/corporeal universe, etc.

If this research is actually valid, I wonder if they are seeing the interface between the non-materialist mind (soul if you like) and the physical brain

The concept of "indeterminacy that is neither random nor deterministic" sounds remarkably vague to me. I mean any signal that is contaminated with noise would fit that description, and attributing magical properties to such a signal would seem nonsensical from a materialist standpoint.

That said, there are situations where adding noise to a weak signal enables it to be amplified more effectively. I don't know much about that subject, but I doubt it is relevant here.

Of course, an ordinary computer can generate pseudo-random noise algorithmically, or it can read quantum-random numbers from a special piece of hardware.

As regards limits to generative AI, I don't think this is relevant at all.

David